Phi-Sat-I: All you need to know about the first AI-based Satellite
Empowered by the AI-enabled chip, the Phi-Sat-1 satellite automatically discards images of earth obscured 70% by the cloud.
Artificial Intelligence is the most appealing technology of the present era. With its ability to mimic human behavior, AI is getting heavily leveraged across diverse sectors for innovative outcomes. Undoubtedly, this eminent technology has enhanced work management by enabling machines to perform redundant and complex tasks for humans. As AI continues to evolve, its use cases are also expanding. Technology and Space Industry have a historical relationship. Since the creation of the first computer, technology is contributing to the breakthroughs of the space industry. Be it in the launch of the first satellite to space or the recent Phi-Sat-1 satellite or humans stepping over the surface of the moon; space missions are heavily governed with technological tools. As the world is witnessing an age of AI, the possibilities of innovations in space get expounded.
In September, an AI-enabled Satellite was launched by the European Space Station in collaboration with chipmaker Intel and a computer vision startup. Cited to be the world’s ‘first AI-Powered Satellite,’ Phi-Sat-1 uses onboard computing and artificial intelligence for advanced satellite imagery. Traditionally, satellite imagery system faces a major challenge of clouds intervention while clicking the space images. The traditional satellite doesn’t have a system to discard skewed images. However, empowered by Intel’s Movidius Myriad 2 chip, the satellite automatically discards images of earth, obscured 70% by the cloud.
The chip gets embedded inside the software created by Ubotica to automatically shut-down in case of overheating. Before the satellite got launched, the AI-based satellite image system underwent a three-week test by taking images across the infrared and thermal-infrared spectrums. Additionally, the traditional satellite processors or hardware doesn’t suit AI-based algorithms. To overcome this challenge, scientists integrated the chip into a particle accelerator at CERN and blasted in for 36 hours. The chip passed two tests and can run multiple applications at the same time.
Apart from discarding obscured cloud images, this AI-enabled satellite will also monitor polar ice, soil moisture, and performing wildfire and flare from oil refineries.
Followed by this breakthrough mission ESA is also looking forward to designing AI-enabled satellite that exceeds satellite imagery capability. For example, to discard the unknown piece of debris in space, ESA has collaborated with a Swiss startup Clearspace, to launch the world’s first debris-removing space-mission, ‘ClearSpace-1 by 2025. The space mission aims to remove a 100 kg payload adaptor orbiting 660 km above the earth. ClearSpace-1 will have an AI-powered camera and robotics arm to grab the debris. The scientist aims to develop a deep learning model to estimate the 6D pose of video sequencing to take pictures of the target. The researchers are planning to install a database of synthetic images for training the algorithms.